The present invention relates to a method for encoding/decoding an error correcting code, a transmitting apparatus and a network which are suitable for use in optical communication networks.
At present, with the advance of digital signal processing technologies based on LSIs and so on, encoding/decoding technologies for error correcting codes have been used in a wide variety of applications for purposes of ensuring a high signal quality. Particularly, among block codes which have a mathematically well-defined organization, a code called “systematic code” is usually used for engineering purposes due to its transparency to information. The systematic code involves segmenting a series of continuous signals into consistent blocks and encoding each of the segmented blocks, and features that only a check bit is added to an empty region, which has been previously determined within the signals, without manipulating information in the original signals. Traditionally, the Hamming code, BCH code (Bose-Chaudhuri-Hocquenghem code), Reed-Solomon code, and so on have been used as block codes. In the following, the encoding/decoding of an error correcting code will be simply called “encoding/decoding.”
The optical fiber communication capable of transmitting a large capacity of data employs relatively high quality transmission paths implemented by optical fibers as media which generally exhibit a bit error ratio below 10−10. In addition, a redundancy configuration, which comprises protection optical fibers as well as working optical fibers, can realize switching of paths when a signal degradation occurs. For this reason, the optical fiber communication has been systematically constructed on the assumption that no error correcting code is used. As a representative example of the optical fiber communication, there is a digital synchronous transmission system for which global standards have been established. This system has been pervasive in transmissions in wide backbone networks all over the world as SDH (Synchronous Digital Hierarchy) defined by International Telecommunication Union (hereinafter called “ITU-T”) in Recommendation G.707 and so on (established in 1988), and SONET (Synchronous Optical Network) defined by American National Standardization Institute (hereinafter called “ANSI”) in Standard T1.105 (established in 1991).
An exceptional introduction of an error correcting code into the optical fiber communication is an application of an eight-error-correcting Reed-Solomon code (255, 239) to a frame format defined by ITU-T in Recommendation G.975 (established in 1996) for a submarine optical transmission system. Also, a known example is JP-A-62-221223.
With the presently widespreading Internet communications, the backbone networks and local networks based on optical fiber communications are required to have the abilities of transmitting increasingly larger capacities of data therethrough. The larger data capacities are being realized by time division multiplexing (TDM), wavelength division multiplexing (WDM), and composite technologies based on them.
However, since a higher degree of time division multiplexing causes a reduced bit width of signals and a degradation in the signal quality resulting from the influence of a variety of dispersion or nonlinearity, which are physical properties inherent to the optical fibers, a certain signal quality can be maintained only over a shorter transmission distance. The optical fiber communication often guarantees a bit rate error of 10−12 or less as the signal quality, and the degree of multiplexing tends to increase with a multiple of two. Also, since the transmittable distance is reciprocally proportional to a square root of the degree of multiplexing for a fixed transmission optical power due to the variance and nonlinearity possessed by an optical fiber, the transmittable distance is reduced to one quarter when the degree of multiplexing becomes twice higher. This reduction corresponds to a degradation loss of 6 dB, so that a compensation for the loss of 6 dB or more is required for increasing the transmission capacity twice as much through the time division multiplexing while the transmission distance is maintained. Thus, for making this compensation for the loss using an error correcting code, a coding gain of 6 dB or more is needed. Since the gain of the eight-error-correcting Reed-Solomon code is 5.4 dB for a bit error ratio of 10−12 in consideration of an increase in the transmission rate by approximately 7%, this error correcting code alone is not sufficient to realize the above-mentioned double increase of the transmission capacity.
Also, as the degree of wavelength division multiplexing becomes higher, this causes closer wavelength intervals of a plurality of optical signals transmitted through a single optical fiber core line, a degraded separation, and a resulting reduction in the transmission distance, similarly to the aforementioned case. In another case, even if the respective wavelength intervals are sufficiently spaced to prevent the degraded separation, the transmission distance is limited when all of bit rates at respective wavelengths are not the same. Specifically, since the transmission distance is determined by the highest bit rate, an optical signal at a low bit rate can be used only within a limited transmission distance although it can be transmitted to more distant locations. The bit rates of a plurality of optical signals transmitted through a single optical fiber core line may differ depending on the generation, the ratio is approximately two in many cases when viewed within a certain period. Therefore, for reasons similar to the aforementioned example, a high bit rate signal must be compensated for a loss of 6 dB or more in order to maximally extend a transmission distance when optical signals at different bit rates are mixed in the wavelength division multiplexed transmission. However, the eight-error-correcting Reed-Solomon code alone is not sufficient to realize such a compensation.
Further, when the distances between regenerators and between a regenerator and an end terminal (hereinafter simply called the “regenerator interval”), for electrically reproducing digital signals, are increased to reduce the number of the regenerators with the intention of reducing the cost associated with the construction of a network at the cost of an increase in the transmission capacity, the signal quality is more degraded as the regenerator interval is longer. For example, when the regenerator interval is increased four times, a compensation for a loss of 6 dB or more is required, in which case the eight-error-correcting Reed-Solomon code alone is not sufficient to realize such a compensation.
Also, the widespreading Internet communications increase a demand for the so-called Giga bits Ether signal of 1000 Base-SX, 1000 Base-LX, 1000 Base-XC defined by IEEE (Institute of Electrical and Electronics Engineers, Inc.) in Standard 802.3z, resulting in requirements for the transmission of the Giga bits Ether signals over a section of a long distance within a local network and a backbone network which accommodate the Giga bits Ether signals as optical signals. Since the Giga bits Ether signal uses a retransmission requesting scheme called ARQ (Auto Repeat Request) based on an end-to-end communication on a higher layer than a link layer, the Giga bits Ether signal comprises no error correcting code.
An error correcting scheme defined in Recommendation G.975 involves parallellizing an STM-16 signal of SDH having a bit rate of 2.48832 Gbit/s on a bit-by-bit basis, dividing the STM-16 signal into (8×n) subframes each having a length of 238 bits, encoding every eight subframes to an eight-error-correcting Read-Solomon code (255, 239), adding a check bit and information for framing structure to the resulting codes, converting the subframes such that each subframe has 255 bits, interleaving the converted (8×n) subframes on a bit-by-bit basis, and finally constructing an FEC frame having a bit rate of approximately 2.666 Gbit/s. In this event, the value of the above “n” is often set to 16 for facilitating the configuration of an encoder and a decoder, in which case, the processing rate is approximately 21 (exactly 19.44×255/238) Mbit/s for each of the subframes.
However, for rearranging the STM-64 signal of SDH, the bit rate of which is 9.95328 Gbit/s, i.e., four times as high as the foregoing, or the OC-192 signal of SONET in the FEC frame, the signal must be divided into four signals corresponding to STM-16 in parallel. This is because the error correcting scheme according to Recommendation G.975 defines the STM-16 signal as a minimum unit. In this event, therefore, the value of the aforementioned “n” is increased by a factor of four from 16 to 64, so that the processing speed in the encoder and the decoder is the same as approximately 21 Mbit/s as mentioned above, where, however, the scale must be increased four times. For example, with the use of encoders and decoders each having the processing capability of approximately 170 Mbit/s, 16 sets are sufficient for the STM-16 signal, whereas 64 sets are required for the STM-64 signal. Also, with the use of encoder/decoders each having the processing capability of approximately 2.7 Gbit/s, one unit is sufficient for the STM-16 signal, whereas four units are required for the STM-64 signal. The increase in the scale is proportional to an increase in the bit rate. For this reason, when a client signal is STM-64 or the like, a codec unit including an encoder and a decoder will be increased in size, resulting in a higher price of a device which contains the codec unit.